PACS photometer map-making with MADmap

1. PACS photometer map-making with MADmap PACS-401 for HIPE 8 user release Version Babar Ali (NHSC)

2. Mapping code written by the Berkeley CMB group to remove 1/f noise from bolometers. http://newscenter.lbl.gov/feature-stories/2010/02/03/madmap/ MADmap was ported to Java for use in HIPE. MADmap offers the so-called optimal map-making to convert time ordered readouts to a final map. Uses maximum likelihood (given a noise/probability model) to determine the optimal sky value. What is MADmap?

3. When should you use MADmap? • When spatially extended emission is present in the data. E.g. Galactic star-formation fields. • MADmap preserves spatially extended emission. • When the source itself is extended. E.g. large galaxies. • When extended structure is present around a compact source. E.g. extended halos or nebulosity.

4. Caveats & Warnings • MADmap assumes that bolometer time-lines are calibrated. • Primarily that pixel-to-pixel instrument variations are already removed. • MADmap assumes that 1/f noise is uncorrelated amongst pixels. • All correlated noise (signal drift) must be removed prior to running MADmap.

5. Masking: bad/dead pixels, glitches, saturation Data Reorganization Logical blocks time stamps Unit conversion Distortion correction Chopper Processing Assign offsets on sky Flag “moving” frames Pipeline section covered here Chopped AOTs Mapping AOTs PACS data, House Keeping Level 0 container Pointing (a) (b) Combine & Subtract ON-OFF beams High Pass Filter Subtract offsets Correct Flat Field Correct Flat Field Correct Flat Field Calibrate Calibrate Initial Processing Flux Calibrate Level 1 cubes Level 1 cube Project on sky Optimal Map Making Algorithm Assign pointing Level 0.5 Minimally processed uncalibrated data cubes Combine individual beams Level 2 images Level 2 images

6. PACS data reduction guide, chapter 9 PACS-101: Introduction to PACS tutorials PACS-103: Accessing & Storing PACS data PACS-104: Using iPipe scripts PACS-201: Level 0 to 1 processing of PACS photometer data PACS-401: MADmap map-making. Documentation Reference

7. Outline of processing • Pre-processing • Calibrate time-lines • Remove correlated signal drift • MADmap • Create “naïve”/simple map. • Create optimal map. • Post-processing (optional) • Remove point-source artifact around bright point sources.

8. Use the tutorials • This talk will concentrate on the using MADmap. • The tutorials provide step-by-step walkthroughs. • Expected results are highlighted. • Recommended parameters and usage is highlighted. PACS 401

9. I: Pre-processing Preparing the time-lines for MADmap

10. Calibrating Time-lines • Start with Level 1 products. • The ipipe scripts start with level 0 • However, the processing is similar between all pipeline branches of the PACS photometer • The following additional steps are needed to finish calibration for MADmap • Assign (a,d) values to each pixel • Remove pixel-to-pixel electronic offsets

11. Taken from the tutorial

12. Check # 5: Offsets are removed Type this command Display( frames.signal[:,:,100] ) Expected Output: With proper offset removal, the image shows a relatively constant signal. Note: extremely bright sources may also be observed on a single image. An example of improper or no pixel-to-pixel offset correction.

13. Masking unstable frames Plot MEDIAN(array) vs Readout # • Unstable frames: • Usually at the beginning • Likely due to calibration pre-amble. To fix:

14. Correlated Signal Drift • Examine the median value of the entire array. • Strong correlation with time (or readout index) • No single model fits the drift (so far. • Behavior is inhomogeneous. Median vs. Time When present, this is the strongest trend in PACS cubes

15. Field guide to signal drifts Typical. Rare. Ascending Hook Strong module-to-module differences

16. Proper correction is crucial Bad Good

17. What to do? • Depending on the complexity of the drift, you may need to iterate with different bin sizes and different orders. • Complicated module-to-module and “hook” drifts require segmenting the timeline in smaller chunks. • Ensure that you plot the medians and the fit (doPlot=True) and you are satisfied with the fit.

18. II. MADmap The actual map-making

19. Step 8 Execute the single line • The ToD stands for Time-ordered-Data and is the internal format used by MADmap. • In fact, makeTodArray will create a binary file in your temporary area that has the rearranged PACS signal in the proper format. • The scale parameter selects the size of the output sky grid relative to the nominal PACS pixel sizes. E.g. scale=0.5 for PACS blue channel will result in final pixel sampling of 1.6”/pixel.

20. Step 9 Select and execute this block of commands • Documentation Reference: • PDRG Chapter 9 • Both the naive and optimal maps are created with the same call. The last parameter is set to ‘True’ for naive map and ‘False’ for optimal map. • MADmap uses maximum likelihood and conjugate gradient solvers to find the optimal solution. The parameters maxRelError and maxIterations control both the convergence tolerance and the number of iterations in finding the optimal solution. See the above reference for details.

21. II. Post-processing Correct the final map for point source artifacts

22. Point-source artifacts Original map Corrected map Artifact map Naïve map

23. How does it work? • Reference: • Lorenzo et al. in prep (available on request) • Developed for HiGal data processing. • Uses the optimal map as the starting point for sky emission. • Subtract sky from original time lines, which leaves only noise+artifacts in the time-line. • High-pass filter the noise+artifacts time-lines. This removes noise and leaves on artifact. • In practice, several iterations are necessary.

24. Step 10 Execute the block of lines The doPGLScorrection flag is set at the beginning of the script. If set, the correctedmap variable will contain the artifact free map. See PDRG Section 9.5 for details. The number of interations for the PGLS algorithm are set in the PGLS_iterations variable (at the start of the script).